On 11/30/2016 10:19 AM, Laurent Pinchart wrote: >> So are we talking about a fork of greybus to do the IoT type stuff >> Pantelis and Alexandre mentioned @ ELCE ? > No, not a fork, a modularization that would make it possible to target a new > field without having to remove UniPro support completely. I kind of implied this in my summary, but I totally agree with this. Nothing should be a fork at this point, it's all evolution of the upstream code to incorporate new or revised functionality. And the right approach is to abstract the UniPro transport so it UniPro isn't strictly required. But I really think there will be more to it than that. I feel like the last year of so we had to really drive SVC protocol to be dependent on the Ara hardware, whereas we've always tried to avoid that. I think we may need to re-think some of that stuff and try to decide what sort of meta-operations (like connection setup and teardown) are really needed to support Greybus's core behavior. And in the process, we *do* need to look at things like (as Jim pointed out) security. Greybus operations rely on a reliable transport, which UniPro could provide. Furthermore, Greybus assumed an enclosed system--a trusted UniPro fabric. There was some partial work on encrypted UniPro connections but that never got done. But once you go beyond that nice constrained system there are lots of secrecy and integrity concerns that Greybus does not address at all. -Alex _______________________________________________ greybus-dev mailing list greybus-dev@xxxxxxxxxxxxxxxx https://lists.linaro.org/mailman/listinfo/greybus-dev
|